Breast tumors require the results of a cytology or pathology examination in order to decide on further management. For non-palpable breast tumors, especially mammography-detectable microcalcifications, the location of the lesion can be determined only from craniocaudal (CC) and mediolateral oblique (MLO) views on mammograms. However, in a case with multiple clusters of microcalcifications or when the image quality is not optimal, it is not easy to localize the microcalcifications using these 2 views. One has to use wire localization to identify the exact site. With this procedure, much manpower and materials must be expended; even the radiation dose of X-rays will be increased due to an increased number of exposures required for needle localization. This is not only hazardous to patients, but also wastes valuable medical resources. To solve this problem, we designed a computer-aided localization system for microcalcifications. After inputting the CC and MLO views of mammograms and with the aid of computer image-processing techniques, a 3-dimensional (3D) breast model was reconstructed to demonstrate and delineate the exact site and distribution of microcalcifications. This model can be used to assist doctors in reducing the number of punctures needed for a biopsy preoperatively, and in demonstrating the exact location of a lesion for reference during operations. The hardware of this system includes equipment necessary to obtain digital images of X-ray mammograms and to network with the personal computer which is also used as an output device to demonstrate the 3D structures of the breast. The flowchart of this software consists of special image- processing techniques of auto-detection and determination of clustered microcalcifications, adjustments for clustered microcalcifications on CC and MLO views, and auto-detection of the nipple, in order to accurately demonstrate the location and distribution of clustered microcalcifications. After delicate testing, the location of the clustered microcalcifications detected from CC and MLO views can be well demonstrated on the constructed 3D model after system calibration, comparison, and axis transformation from the designed software. The exact location of the lesion can be viewed with a virtual reality modeling language viewer.
|Number of pages||12|
|Journal||Chinese Journal of Radiology|
|Publication status||Published - 2001 Jan 1|
All Science Journal Classification (ASJC) codes
- Radiology Nuclear Medicine and imaging